Computer-aided segmentation and estimation of indices in brain CT scans

The importance of neuro-imaging as one of the biomarkers for diagnosis and prognosis of pathologies and traumatic cases is well established. Doctors routinely perform linear measurements on neuro-images to ascertain severity and extent of the pathology or trauma from significant anatomical changes. However, it is a tedious and time consuming process and manually assessing and reporting on large volume of data is fraught with errors and variation. In this paper we present a novel technique for segmentation of significant anatomical landmarks using artificial neural networks and estimation of various ratios and indices performed on brain CT scans. The proposed method is efficient and robust in detecting and measuring sizes of anatomical structures on non-contrast CT scans and has been evaluated on images from subjects with ages between 5 to 85 years. Results show that our method has average ICC of ≥0.97 and, hence, can be used in processing data for further use in research and clinical environment.